Citation Context

...rity [35]. Bloom filters [7] are commonly used as compact representations of large sets. The R* distributed database [33] calculates them on demand to implement efficient semijoins. The Summary Cache =-=[16]-=- pushes Bloom filters between cooperating web caches, although their method does not scale well in the number of caches. Distributing data for performance, availability, or survivability has been stud...

Citation Context

...servers. CDNs are typically managed by a central entity, while CFS is built from resources shared and owned by a cooperative group of users. There are several proposed scalable cooperative Web caches =-=[3, 8, 10, 15]-=-. To locate data, these systems either multicast queries or require that some or all servers know about all other servers. As a result, none of the proposed methods is both highly scalable and robust....

Citation Context

...ifier associated with each message. Each host stores a bit vector, called the summary vector that indicates which entries in their local hash tables are set. While not explored here, a “Bloom filter=-=” [4, 12]-=- would substantially reduce the space overhead associated with the summary vector. When two hosts come into communication range of one another, the host with the smaller identifier initiates an anti-e...

Citation Context

...asy to see that if the query returns false, then y is not in the set, while if it returns true, then it might be. Classical Bloom filters do not permit elements to be removed. A counting Bloom filter =-=[6]-=- replaces the Boolean array B with an array of counters C. Adding an element x (atomically) increments each C[hi(x)], and removing the element (atomically) decrements each such counter. A member(x) qu...

Citation Context

...can also serve to decrease object access latencies. Much research has focused on Web proxy caching [4, 5, 7, 11, 12] and, more recently, on coordinating state among multiple, cooperating proxy caches =-=[13, 30, 33]-=-; some of these proposals aim to create global caching structures [27, 34]. The results of these studies generally indicate that cache hit rates of 40-50% are achievable, but that hit rate increases o...

Citation Context

...g systems fit within a large class of CDNs that handle high demand through diverse replication. Prior to the recent interest in peer-to-peer systems, several projects proposed cooperative Web caching =-=[2, 7, 8, 16]-=-. These systems either multicast queries or require that caches know some or all other servers, which worsens their scalability, fault-tolerance, and susceptibility to hot spots. Although the cache hi...

Citation Context

...ents across organizations, the ratio of inter-proxy communication time to server fetch time, and the scale at which cooperation is undertaken. Several cooperative-caching protocols have been proposed =-=[9, 16, 28, 30, 33]-=-; however, few studies have examined cooperative Web caching from a systemic viewpoint. As a result, we know neither the environments in which cooperative caching is useful (if any) nor its potential ...

Citation Context

...g them, and each cache server becomes the designed server for a number of Web sites. User requests are then forwarded to the proper cache server according to the cache routing table. In Summary Cache =-=[29]-=-, Cache Digest [68], and the Relais project [66], caches interexchange messages indicating their content and keep local directories to facilitate finding documents in other caches. 4.1.3 Hybrid cachin...